Robust Ante-hoc Graph Explainer using Bilevel Optimization

TL;DR

This paper introduces RAGE, a novel ante-hoc graph explainer using bilevel optimization, which identifies informative molecular substructures for graph neural networks, improving explanation quality in chemical applications.

Contribution

The paper presents RAGE, a new ante-hoc explainer that leverages bilevel optimization to produce more informative and relevant explanations for graph neural networks, especially in chemistry.

Findings

RAGE outperforms existing explainers in molecular classification tasks.

RAGE effectively identifies molecular substructures containing full predictive information.

RAGE explanations are more informative and relevant than prior methods.

Abstract

Explaining the decisions made by machine learning models for high-stakes applications is critical for increasing transparency and guiding improvements to these decisions. This is particularly true in the case of models for graphs, where decisions often depend on complex patterns combining rich structural and attribute data. While recent work has focused on designing so-called post-hoc explainers, the broader question of what constitutes a good explanation remains open. One intuitive property is that explanations should be sufficiently informative to reproduce the predictions given the data. In other words, a good explainer can be repurposed as a predictor. Post-hoc explainers do not achieve this goal as their explanations are highly dependent on fixed model parameters (e.g., learned GNN weights). To address this challenge, we propose RAGE (Robust Ante-hoc Graph Explainer), a novel and flexible ante-hoc explainer designed to discover explanations for graph neural networks using bilevel optimization, with a focus on the chemical domain. RAGE can effectively identify molecular substructures that contain the full information needed for prediction while enabling users to rank these explanations in terms of relevance. Our experiments on various molecular classification tasks show that RAGE explanations are better than existing post-hoc and ante-hoc approaches.